Molecular orbital analysis of the orientation-dependent barrier to direct

Dimers of and Tautomerism between 2-Pyrimidinethiol and 2(1H)-Pyrimidinethione: A Density Functional Theory (DFT) Study. Fillmore Freeman , Henry N. P...
0 downloads 0 Views 1MB Size
VOLUME 113, NUMBER 9 APRIL24, 1991 Q Copyright 1991 by the American Chemical Soeiery

TOU R N A L OF THE AMERICAN /

CHEMICAL SOCIETY

Molecular Orbital Analysis of the Orientation-Dependent Barrier to Direct Exchange Reactions Davide M. Proserpio,+Roald Hoffmann,*vt and R. D. Levine*?* Contribution from the Department of Chemistry and Materials Science Center, Cornell University, Ithaca, New York 14853- 1301, The Fritz Haber Research Center for Molecular Dynamics, The Hebrew University, Jerusalem 91 904, Israel, and the Department of Chemistry and Biochemistry, University of California, Los Angeles. California 90024- 1569. Received June 11, 1990

Abstract: After a general introduction to the current state of potential energy functions for delineating the steric requirements of exchange reactions, we proceed to a qualitative, Walsh-diagram-based orbital analysis of the preferred geometry approach to the barrier to such reactions. Extended Huckel calculations, perturbation theory, and a frontier orbital perspective are used to analyze the X + H2 (collinear, due to the controlling role of the singly occupied highest molecular orbital), X + HX, X + HY (also collinear), H + X2 (bent, due to interaction with a relatively low lying unfilled orbital), H + XY (a preference for direction of attack depending on relative electronegativities), M + X,, M + XY, M + HX (collinear approach slightly preferred), X + XY, and triplet 0 atom reactions.

Introduction Concerted atom exchange reactions, even when exothermic, do typically have an activation barrier. The rate of such reactions is often lower that the collision rate computed for hard spheres with enough energy to surmount this barrier.’ The venerable collision theory of chemical kinetics introduces therefore an ad hoc “steric factor” as the ratio of the observed to computed rates. The origin of the correction was clearly understood: the reaction has not only energetic but also steric requirements. However, no prescription for computing this factor was provided. An early success of transition-state theory1 was that it produced a quantitative account of the steric factor. In terms of the thermodynamic formulation of the theory, the steric factor is due to the entropy loss in going from the free rotation in the reagents to the very restricted rotation in the transition state. Transition-state theory by its very nature deals with thermal reactants. The direct experimental demonstration of the steric requirements of exchange reactions2s3 requires the collision of oriented reactants. Indeed, in the entire area of molecular reaction dynamics: progress has been due to the use of selected reagents ‘Cornell University. *The Hebrew Uniiersity.

which are necessarily not in equilibrium. Exact dynamical computations (whether classical or quantum) can, of course, begin with a sharply defined initial state and hence are directly relevant to scattering experiments. Moreover, with proper averaging over the initial conditions they determine the magnitude of the steric factor. Such computations require as input the potential energy function (which varies with the interatomic distances) of the system^.^^^ In principle, the potential energy function can be determined by the standard methods of ab initio quantum chemistry.’-I0 In (1) Laidler, K. J. Chemical Kinetics; Harper and Row: New York, 1987. (2) Brooks, P. R. Science 1976,193, 11. Stolte, S. Ber. Bunsenges. Phys. Chem. 1982, 86,413. Bernstein, R. B.; Herschbach, D. R.; Levine, R. D. J . Phys. Chem. 1987, 91, 5365. (3) Bernstein, R. B. Chemical Dynamics via Molecular Beam and Luser Techniques; Oxford University Press: New York, 1982. (4) Levine, R. D.; Bernstein, R. B. Molecular Reaction Dynamics and Chemical Reactiuity; Oxford University Press: New York, 1987. ( 5 ) Hirst, D. M. Potential Energy Surfaces; Taylor and Francis: London, 1985. Truhlar, D. G., Ed. Potential Energy Surfaces and Dynamics Calculations; Plenum: New York, 198 1. (6) Schatz, G. C. Rev. Mod. Phys. 1989, 61, 669. (7) Truhlar, D. G.; Steckler, R.; Gordon, M. S . Chem. Rev. 1987,87,217. (8) Mezey, P. G. Potential Energy Hypersurfaces; Elsevier: Amsterdam, 1987.

0002-7863/91/1513-3217%02.50/00 1991 American Chemical Society

Proserpio et al.

3218 J . Am. Chem. SOC.,Vol. 113, No. 9, 1991 practice this has only been implemented, for a sufficiently wide range of interatomic distances, for a rather limited set of reactions. Potential energy functions for use as input for (exact or approximate) dynamical computations are currently being generated primarily by a valence bond semiempirical approach. The most common procedure, knows as LEPS,5 is based on the zeroth order, London-type approach which is strictly valid only for S-state atoms. An LEPS-type function will usually predict a collinear transition state. More recently, a refined, diatomic-in molecules (known as DIM) procedure has been extensively applied.11-’4 A large data set is required as input for the semiempirical DIM procedure which does, however, lead to results often qualitatively different from that of the simple LEPS limit. In particular we note that it often favors a not strictly collinear approach. For stable ABC molecules, Walsh15 considered the variation of the molecular orbital energies with bend angle. In this fashion he was able to correlate the equilibrium geometry of the molecule with the number of valence electrons. The Walsh procedure can equally well be used to examine the bend angle for which the ABC transition state in the A BC reaction has the lowest energy. A qualitative molecular orbital interpretation of the steric requirements of atom exchange reactions was thereby provided.Ibl8 In this paper we take the molecular orbital approach one step further by using computed, rather than estimated, molecular orbital energies. The computations are performed within the extended Hiickel f0rma1ism.I~ This is a very approximate M O approach with real limitations for geometrical optimization. It does, however, seem to get the bending of the molecule right. Moreover, it has the advantage of transparency in that it provides a very direct, perturbation theory based on connection to simple orbital concepts. In this way it retains the advantage of qualitative methods of correlating the geometry of the transition state to orbital occupancy in the reagents. Strictly speaking we need to compute the entire potential energy function. In this preliminary attempt we pursue a more modest goal, justified by the following consideration: To a zeroth approximation one can continue to adopt for the dynamics the old collision theory point of view with one essential change.4-2*22 To account for the steric requirements it is necessary to regard the barrier to reaction as depending on the orientation of the reactants (that is, on the ABC bend angle for the A BC reaction). The object of our computation is to determine this dependence. The concept of an orientation-dependent barrier to reaction is an important one because it allows “inversion” from experimental data. In other words, dynamical ~tereochemistry~~ experiments can provide a rather direct handle for a key feature of the potential energy function. Such an inversion has indeed been demon~trated.~~-~~ Already at the level of the orientation-dependent barrier to reaction there can be a meeting of theory and experiment. Other aspects of dynamical stereochemistry can be found in two recent conference proceeding^^^*^^ and, in general, in the reports of any

+

+

(9) Lawley, K. P., Ed. Ab initio Methods in Quantum Chemistry; Wiley: New York, 1987. (10) Dykstra, C. E.Ab initio Calculation of the Structures and Properties of Molecules; Elsevier: Amsterdam, 1988. (11) Tully, J. C. Adv. Chem. Phys. 1980, 42, 63. (12) Kuntz, P. J. Ber. Bunsenges. Phys. Chem. 1982, 86, 413. (13) Duggan, J. J.; Grice, R. J . Chem. Phys. 1983, 78, 3842. (14) Last, I.; Baer, M. J . Chem. Phys. 1984, 80, 3246. (15) Walsh, A. D. J . Chem. Soc. 1953, 2260, 2266,2288. (16) Herschbach, D. R. Adu. Chem. Phys. 1966, 10, 319. (17) McDonald, J. D.; Le Breton, P. R.;Lee, Y.T.; Herschbach, D. R. J . Chem. Phys. 1972,56,769. Carter, C . F.; Levy, M. R.;Grice, R. Faraday Discuss. Chem. Soc. 1973, 55, 357. (18) Mahan, B. H. Acc. Chem. Res. 1975,8, 5 5 . (19) Hoffmann, R.; Lipscomb, W. N. J . Chem. Phys. 1962, 36, 2179, 3489. Hoffmann, R. Ibid. 1963, 39, 1397. (20) Smith, I. W. M.J . Chem. Educ. 1982, 59, 9. (21) Jellinek, J.; Pollak, E. J . Chem. Phys. 1983, 78, 3014. (22) Levine, R. D.; Bernstein, R. B. Chem. Phys. Lett. 1984, 105, 467. (23) Special issue: J . Phys. Chem. 1987, 91, (21). (24) Bernstein, R. B. J . Chem. Phys. 1985, 82, 3656. (25) Loesch, H. J.; Hoffmeister J . Chem. SOC.,Faraday Trans. 2 1989, 85, 1249. (26) Special issue: J. Chem. Soc., Faraday Trans. 2 1989,85, (8).

recent conference dealing with the dynamics of molecular collisions (e.g. refs 27-29). The available experimental and computational results show that such exchange reactions which proceed without an activation barrier (e.g. certain ion-molecule reaction^^-^^-") have their steric requirements dominated by the anisotropy of the long-range physical forces. Here, however, we are concerned with activated atom exchange for which the orientation dependence of the chemical barrier is the relevant factor. We note here a distinction between two basic ways in which orientation effects can be manifested in the A + BC reaction. The first, the one of concern here, is the orientation dependence of the barrier to reaction which governs the reactivity. The second regards the dynamics of the trajectories of real molecules as they traverse the potential energy surface in their approach motion to the barrier. The anisotropy of the long-range physical forces and, at the shorter ranges, of the chemical forces en route to the barrier can orient the BC molecule during the approach of A. Time scales of motion are then important as this reorientation is particularly significant for lower velocity collisions and/or if the longer range force is very anisotropic. It is of relevance in understanding the reaction of initially oriented reactants (whether the initial orientation is achieved in the gas phaseg2or via using a condensed medium33). It is also of prime importance in understanding the reaction when BC is initially rotationally e x ~ i t e d . ~Irrespective, ~.~~ however, of whether BC did or did not significantly reorient during the approach motion to the barrier, reactivity is determined by the orientation dependence of the barrier itself. When the BC bond is significantly stretched, the orientation dependence of the barrier can change.36 In extreme cases even new reaction mechanisms can become feasible. An example is the H H2 reaction which ordinarily proceeds via collinearly dominated abstraction. At collision energy above 20 kcal mol-I or so, a new insertion-type mechanism, with a sideways approach, becomes feasible.” For vibrationally excited (e.g. laser pumped) reagents the cone of acceptance typically opens up.3* Here we shall not explore such dependence in detail. In practice we have, however, varied the BC bond distance so as to verify that our results do not change in a significant way when the BC bond is somewhat stretched. Finally we note that the transparency of the extended Hiickel procedure and its direct tie to a frontier orbital perspective make it useful in another way. Even if the complete surface is not reliable, identification of the primary interactions (Le. which orbitals govern a linear or nonlinear geometry) could lead to the optimum functional form for describing a surface. Better calculations could be fitted to the functional form and used in dynamical studies.

+

The Orientation-DependentBarrier to Reaction A simple interpretation of the bonding at the barrier for reaction can be obtained by considering the occupancy of the molecular orbitals for the ABC ~ y s t e m . ~ *Following I ~ ~ * the mode of presentation introduced by Walsh15 in the study of the geometry of stable ABC molecules, we consider the orbital energies vs the ABC bend angle. (27) Whitehead, J. C., Ed. Selectivity in Chemical Reactions; Kluwer: Dordrecht, 1987. (28) Faraday Discuss. Chem. Soc. 1987, 84. (29) Vetter, R.; Vigue, J., Eds. Recent Advances in Molecular Reaction Dynamics; CNRS: Paris, 1986. (30) Clary, D. C. Mol. Phys. 1984, 53, 3. (31) Levine, R. D.; Bernstein, R. B. J . Phys. Chem. 1988, 92, 6954. (32) Bernstein, R. B.; Levine, R. D. J . Phys. Chem. 1989, 93, 1687. Loesch, H. J. Chem. Phys. 1986, 104, 213. (33) Benjamin, I.; Liu, A.; Wilson, K. R.; Levine, R. D. J . Phys. Chem. 1990, 94, 3937. (34) Loesch, H. J. J . Chem. Phys. 1987, 112, 85. (35) Komweitz, H.; Persky, A,; Levine, R. D. Chem. Phys. Lett. 1986,128, 443. (36) Schechter, I.; Kwloff, R.; Levine, R. D. Chem. Phys. Lett. 1985, 121, 297. (37) Schechter, I.; Levine, R. D. Int. J . Chem. Kinet. 1986, 18, 1023. (38) Schechter, I.; Kosloff, R.; Levine, R. D. J . Phys. Chem. 1986, 90, 1006.

J . Am. Chem. SOC.,Vol. 113, No. 9, 1991 3219

Orientation- Dependent Barrier to Direct Exchange Reactions Table 1. Atomic Parameters Used in the Calculations

atom

F CI Br

I Li Na

K Rb

0 H

orbital 2s 2P 3s 3P 4s 4P 5s 5P 2s 2P 3s 3P 4s 4P 5s 5P 2s 2P Is

H,,(eV)

fl

-40.0 -18.1 -26.3 -14.2 -22.07 -1 3.1 -18.0 -1 2.7 -5.4 -3.5 -5.1 -3.0 -4.34 -2.73 -4.18 -2.6 -32.3 -14.8 -13.6

2.425 2.425 2.183 1.733 2.588 2.131 2.679 2.322 0.650 0.650 0.733 0.733 0.874 0.874 0.997 0.997 2.275 2.275 1.30

,*

~

(A) 0.71

4 0 (LUMO)

0.99 1.14 1.33 30 (SOMO)

1.52 H-H

1.86 2.27

H-H

'

1%

=Y

2.47 0.60 0.37

The calculations are performed with the extended Huckel methcd19 with weighted In the spirit of choosing a location on the surface characteristic of the onset of chemical forces, a distance is selected at which significant orbital interaction between the approaching atom and the diatomic molecule occurs. This is done in the following way: we calculate the overlap population (OP) between A and BC for different distances (6)in the linear geometry. A separation yielding an OP between 0.09 and 0.1 1 is chosen, and then it is fixed as angular variations are made. When this value is impossible to reach, we look at the OP for B-c, compared to the value for the isolated molecule (OP,). We select the distance when the difference (OP, - OP) is between 0.09 and 0.1 I . Using these criteria we found in general that d = [rl + r2 0.4 ( I ) A], where rl and r2 are the covalent radii of the adjacent atoms. The bond distance of the reactant diatomic is kept at its value for the isolated molecule. We did, however, verify that our conclusions do not qualitatively change under small variation either of the value of d and/or of the diatomic bond distance. Covalent radii and other parameters used in the calculations are standard literature values and are collected in Table I. The three-dimensional graphics have been carried out by a computer program named CACAO, described elsewhere.40 For each reaction we show (i) the molecular orbital interaction diagram including the atomic orbital used and (ii) the computed Walsh diagram, with the shape of selected orbitals. Where available, we compare data with previous a b initio and/or semiempirical calculations and with experimental insights. The results are discussed separately for each family of reactions. X and Y will denote a halogen atom and M a metal atom. The X + H2 Reaction. Consider first the very well studied F + H2r e a c t i ~ n . ~Both , ~ ~ the ab initio computation^^^ and the best semiempirical LEPS-type surface43concur that the lowest energy barrier is for a collinear configuration. The recently noted apparent disagreement of dynamical computations using this surface with experimentM had led to a purely empirical modification of the LEPS function which predicts a flat bent potential with a noncollinear lowest saddle point.4s This does seem to improve the agreement with the experiment and there is a preliminary ab initio computation4 showing that higher order correlation effects

2u

Figure 1. Selected molecular orbitals for the reaction F + H-H in the linear approach.

lead to the lowering of the bending potential. The orbital interaction diagram for F H2 is shown in I. The labeling of the molecular orbitals correspond to both the linear (C-") and bent (C,)symmetry.

+

P. Py P.

+

(39) Ammeter, J. H.; Biirgi, H.-8.; Thibeault, J. C.; Hoffmann, R. J. Am. Chem. Soc. 1978, 100, 3686. (40) Mealli, c.;Proserpio, D. M. J . Chem. Educ. 1990, 66, 399. (41) Steinfeld, J. 1.; Francisco, J. S.;Hase, W. L. Chemical Kinetics and Dynamics; Harper and Row: New York, 1989. (42) Schaefer, H. F., 111 J . Phys. Chem. 1985,89, 5366. (43) Muckermann, J. T. In Theoretical Chemistry; Henderson, D., Ed.; Academic Press: New York, 1981. (44) "nark, D. M.; Woodtke, A. M.; Robinson, G. N.; Hayden, C. C.; Lee, Y. T. J . Chem. Phys. 1985,82, 3048. (45) Takayanagi, T.; Sato, S.Chem. Phys. k r r . 1988,144, 191. See also Pattengil et al. (Pattengil, M. D.; Zare, R. N.; Jaffe, R. L. J . Phys. Chem. 1987, 91, 5498) for a different possible empirical modification of the LEPS functional form.

2a + 2a'

F

F

.....H-H I5A

H-H

0.74A 1

There are a total of nine valence electrons in this system, of which seven are shown in I. The 1a orbital, almost entirely F 2, is omitted. It should be noted explicitly here that our MO numbering system is designed for valence-electron calculations. In all electron computations the F 1s would be la, and the 2s 2a and what we call 20 would be labeled 3a. The highest populated molecular orbital, which may be described as an out-of-phase combination of the ag bonding MO of Hz and the fluorine pz orbital (see molecular orbital 3a in Figure l), is singly occupied. We call this critical orbital SOMO (singly occupied molecular orbital). The SOMO 4a' is an antibonding orbital, and the increase in energy due to the occupancy of this orbital is largely responsible for the (small) activation barrier for the reaction. This suggests that the dynamics of the ion-molecule F+ + H2reaction4' would be of interest, except for possible complexity due to nonadiabatic coupling with the F H2+ channel. We explore nonlinear approaches of F by varying the angle 0 from 180 to 90°, keeping all the distances constant (11).

+

F

..-..

e

I1

The calculated Walsh diagram is shown in Figure 2 (top). The dashed line on the diagram is the total energy of the system on a scale which is the same as that of the orbital energies. On the bottom we illustrate the shape of the 4a'SOMO along the six studied steps of the distortion. The increase of antibonding re(46) Schwenke, D. W.; Steckler, R.; Brown, F. B.; Truhlar, D. G. J. Chem. Phys. 1986.84, 5706. (47) Stowe, G.; Armentrout, P. B. Private communication.

3220 J . Am. Chem. SOC..Vol. 113, No. 9, 1991

Proserpio et al.

@ --, ,,..., ...*. 6

*--a*. *a

*-..o-.

1h?"

IW*

H@

30

144"

Figure 3. Selected molecular orbitals for the reaction F linear approach.

+ H-F

in the

The orbitals here are closely related to those of a symmetric F-H-F, a system well-explored in the anion case. In the centrosymmetric molecule we would have the following level structure (IV). Note how the orbitals of Figure 3 are small perturbations of these. Figure 2. Walsh diagram for the reaction F + H-H (top), with the shape o f the SOU0 4a' along the distortion (bottom).

pulsion between fluorine and the two hydrogens (which is what originally destabilized this combination) explains the rise in energy of this level, and with it of the total energy. The collinear approach is then preferred. The near-parabolic increase of the 4a' level with bending angle vs the linear decrease of the 3a' and 2a' molecular orbitals suggests the in principle existence of a possible bent transition state, provided that the energy of the latter MO's decreases with the bending angle faster than we find in the computation. Our results for the other halogens (vide infra) suggest that such would be the case for an atom even more electronegative than fluorine (e.g. Ye'). Substitution of fluorine by other halogens does not change our qualitative results. The less electronegativeatom (CI, Br, I) shifts the np orbitals to higher energy. The more diffuse p orbitals have, at the same distance, stronger overlap and consequently stronger antibonding interaction with H2 crg in the SOMO. The collinear approach is always preferred and more so the heavier the halogen. The X + HX Reaction. For the reaction F + HF, the SOMO 6a' is a combination of two fluorine pz orbitals, with almost no contribution from the hydrogen. There is also a greater mixing of the incoming F (see 111 and Figure 3). 5a + 7a' 1.1;510

PI

Py

OIIF*

20

4

-e-

l \.

The Walsh diagram for the approach of the fluorine at different 8 is shown in Figure 4. The shape of the SOMO shows (Figure 4, bottom) that the lobe of the p orbital of the fluorine in H F is deformed by mixing the H 1s in a bonding way in the wave functions. The interaction with the incoming F is more antibonding. The total energy rises and therefore the collinear approach is preferred. Substitution of fluorine by heavier halogens has the same effect as in X + H-H. The more diffuse orbitals interact strongly in the SOMO. The collinear geometry is the more stable (see for example Figure 5 for CI + H-CI). We can compare our qualitative results with the calculated potential barrier reported by Last and Baer14 for different geometries:

PI

react ion

F + HF

CI + HCI Br + HBr I + HI F

..... H-F I.VA 0.9rA

111

H-F

AE (eV) = E(9OO) - E( 180') 0.45 I .28 I .37 I .42

These results confirm that the collinear approach is always preferred and this effect increases as one moves down group 17. For other DIM computation including the role of spin-orbit interaction see ref 48. Ab initio results for this reaction, discussed

J. Am. Chem. SOC.,Vol. 113, No. 9, 1991 3221

Orientation- Dependent Barrier to Direct Exchange Reactions

rl

-- 79

1800

90'

FHCl Bend Coordinate (81

*--a

*... & 180'

:

'a

t

b-a

'3.

1269

90'

180"

FHF Bend Coordinate (€9

*$

n

1

-9 -7

6a'

90"

103'

Figure 4. Walsh diagram for the reaction F + H-F (top), with the shape of the SOMO 6a' along the distortion (bottom).

162"

:

*
5 A) one electron is transferred from the metal to the molecule. This jump is governed by the small difference between the ionization potential of the metal and the electron affinity of the molecule. At short range, where orbital interaction effects are important, the governing surface can be thought of as that of M+ X2-.The orbitals of the metal should now be less diffuse. We simulated this by increasing the exponent of the Slater function to a mean value of l.l.67 To effect the electron transfer we

+

+

+

(64) Jorgensen, W. L.; Salem, L. The Organic Chemist's Book of Orbitals; Academic Press: New York, 1974. (65) Our calculations thus give substantial electron density on the central atom. This disagrees with conclusions based on calculations by D. A. Dixon for the A + BC system, cited in: Truhlar, D. G.; Dixon, D. A. AromicMolecule Collision Theory; Bernstein, R. B., Ed.; Plenum: New York. 1979; Chapter 18. (66) Davidovits, P.; McFadden, D., Eds. The Alkali Halide Vapors; Academic Press: New York, 1979.

Proserpio et al.

3224 J . Am. Chem. SOC.,Vol. 113, No. 9, 1991 1

- 57

+

-..I

I

I

180"

900

NaClCl Bend Coordinate (8)

For the reaction M XY, there is no preference for attack on one or the other of the two halogens, the interaction with the uu of the elongated molecule being approximately equal for both sides. We find that the linear approach is slightly preferred in either case. We reach the same conclusion when M is substituted by metals of group 2. The 6a' level is now doubly occupied, and the linear approach is preferred. The M + X H Reaction. For the hydrogen halides the longrange harpoon mechanism is no longer valid. The transfer of one electron could occur only at small separation where the orbital interaction effects are important. For this reaction we adopt the usual approach, keeping the HX distance at the equilibrium value. The diagram is similar to X,but the interaction between the metal s A 0 and the empty antibonding a* orbital of XH is now weaker. The frontier orbitals do not change for all the possible reactions (e.g. Li + FH, Na + IH), and the Walsh diagrams do not vary much. The total energy increases slightly from 180 to 90°, but by less thari 0.1 eV. It seems a wide range of approach angles is available for this class of reactions. We can improve the charge transfer from the metal to the molecule, as we did before, by elongating the H-X bond so as to lower the u, level. The result is a preferred linear geometry as for M + X2. Our results are more in accord with DIM computation~~' than with the bent configuration predicted by ab initio method^^^-^^ for both Li + HF and Li + HCI, with the stretched H-X bond at the transition state. As a reviewer has pointed out, the saddle points for the M XH reaction are in the exit channel, a stable Me-X-H intermediate being formed. It may be that in this case our calculations, which probe the beginnings of the exchange reaction, i.e. the entrance channel, are not relevant. The X X2,Y X2,and X + XY Reactions. The interaction diagram for X + X2 is shown in XI. The system has a total of 21 electrons, with the SOMO an antibonding combination between the incoming X(p,) and the ug of X2.

+

'a

'a

126'

108"

5. 90"

Figure 10. Walsh diagram for the reaction Na + CI-CI (top), with the shape of the SOMO 6a' along the distortion (bottom).

elongated the X2 distance, lowering the energy of the X2 u,. This is the acceptor orbital for the electron transferred from M. For our computation we choose the approach distance so that for the collinear reaction the charge on the metal is near +0.8 and the OP (M.-X2) near 0.1. The interaction diagram X shows that the SOMO is the bonding combination of the u, from X2 and the s orbital from bf.

+

+

377 -+ 3a"+ 7a'

P. Py P* 2n -+ 2a"+ 6a'

lx -+ la"+ 5a' PI PI

-

Py P. s

371 -+ 3a"+ &

*

G

a

E:

40 -+ 4a'

x . .. .x-x

u Eki -+ 7a' a"

-

x-x

XI

In the Walsh diagram for the Br + Br2 reaction (see Figure 1l ) , the levels 8a' (SOMO) and 7a' are strongly affected by the Idistortion. The preferred geometry depends on the energy balance between In -+ la"+ 4a' the doubly occupied 6a' and the singly occupied 7a'. Our results =a shown that the total energy curve is flat for F F2and CI Clz, aE' r 3a -+ 3a' with a shallow minimum at 145' (ca. 0.15 eV) for Br Br2, M . . . .X-X x-x decreasing linearly to 90' (ca. 0.37 eV) for I 12. The DIM results75show the bending already for C1+ Clz. A system with X 22 electrons (X-+ X2) will prefer the linear approach. Now we consider as an example the Na + C12 reaction, with We also consider the reaction between different halogens, Y the Walsh diagram in Figure 10. The energy of the SOMO 6a' + X2. The relative position of the FMO's interacting in XI is increases along the distortion. The contribution of the metal s different, but the nature of the SOMO, the antibonding combiA 0 in the molecular orbital decreases. The collinear approach nation between pz and ug,is always 8a' in all possible reactions is slightly preferred, for all possible reactions, by 0.2-0.4 eV. This from F + I2 to I + F2. The bent approach (ca. 145') is most is in accord with DIM computations6* and with indications prostable. with a relative minimum of 0.2-0.4 eV. vided by a b initio computation^.^^^^^ 4a -+ 6a'

271 -+ 2a"+ Sa'

+

+

(67) For a STO the exponent ( is proportional to Z*/n. Increasing the charge by one we obtain = (f + l/n). The mean value for the alkali metals is 1 . 1 . (68) Zeiri, Y.;Shapiro, M. J . Chem. Phys. 1981, 75, 1170. (69) Balint-Kurti. G. G. Mol. Phys. 1973, 25, 393. (70) Maessen, B.; Cade, P. E. J . Chem. Phys. 1984, 80, 2618.

r

+ +

(71) Shapiro, M.; Zeiri, Y . J . Chem. Phys. 1979, 70, 5264. (72) Chen, M. M. L.; Schaefer, H. F.,111 J . Chem. fhys. 1980, 72,4376. (73) Garcia, E.; Lagani, A,; Palmieri, A. Chem. Phys. Left. 1986, 127, 73. (74) Palmieri, A.; Garcia, E.; Lagani, A. J . Chem. Phys. 1988.88, 181. (75) Duggan, J. J.; Grice, R. J . Chem. Soc., Furuduy Truns. 2 1984,80, 795, 809.

Orientation- Dependent Barrier to Direct Exchange Reactions

1800

J . Am. Chem. SOC.,Vol. 113, No. 9, 1991

3225

reaction we can use the interaction diagram in VI. Because oxygen is in a jP state we remove one electron from the inner level, doubly degenerate, under the SOMO. The final state with 14 electrons is now a triplet that correlates with the reactants. The resulting Walsh diagram for the distorted approach is similar to that in Figure 7 and so is the total energy. The triplet state prefers the linear approach, and the rapid barrier rise for angles below 145' or so is as assumed in the empirically optimized LEPS surface.77 The most recent ab initio results78 suggest however a shallow off-collinear minimum. In the same way we can examine O(3P) + XY looking at XI and Figure 11. We are making the assumption, supported by actual calculations, that the qualitative features of the X + XX diagram carry over to Z + XY. For the same overlap reason as before, the attack on less electronegative halogen is preferred. The electron is removed from one of the inner levels that transforms as 6a' and 2a" in C,symmetry. These levels are unaffected by the bending, and as a consequence the total energy depends again on the balance between the SOMO 8a' and the 7a'. We found that the bent (ca. 145') approach is stabilized.

90'

BrBrBr Bend Coordinate (8;

Figure 11. Walsh diagram for the reaction Br + Br-Br (top), with the shape of the SOMO 8a' (center) and the level 7a' (bottom) along the distortion.

The most general set of reactions Z + XY, with three different halogens, and X + XY does not present new features. As for H + XY, the attack on the less electronegative atom is energetically favored. The Walsh diagram for the bent reaction is very similar to the one in Figure 11, the approach on the side being slightly stabilized in all cases examined. The position of the minimum changes from 145 to 90' for different combinations of halogens, but the stabilization is never more than 0.6 eV. These trends are consistent with the early experimental results.76 The O(jP) + XH and O(jP) XY Reactions. In all reactions examined the intermediate complex (radical) usually corresponds to a molecular species which correlates with the ground state of reactants and products. In the reaction with oxygen, however, different spin states are involved. To analyze the O(jP) H X

+

+

(76) Lee, Y.T.; LeBreton, P. R.; McDonald, J. D.; Herschbach, D. R. J . Chem. Phys. 1969, 51, 455.

Concluding Remarks and Prospects The results reported show that orbital considerations and their quantitative expression as Walsh diagrams can be used to determine the atomic orientation near the barrier for exchange reactions. Even if not quantitatively accurate, the extended Hiickel approach captures the chemical systematics when s and p orbitals are important. Since there are clear reactivity trends with d orbital o c c ~ p a n c y , ~it~would * * ~ be desirable to apply such considerations to them as well. Another important direction is to determine the entire potential energy function. In particular, there is a clear need to extend the LEPS functional form with its collinearly preferred approach to an analytical form valid also when p orbitals are involved. Molecular orbital considerations clearly establish why a collinear approach is expected for s orbitals. The present results show that the method can discern between collinear and sideways approach when p orbitals are occupied. It can even account for variations within a given family (as in X + X2, from collinear for F + F2 to completely bent for I 12). It remains, however, to provide an analytical representation of such systematics.

+

Acknowledgment. This work began when one of us (R.D.L.) visited Cornel1 University. He would like to thank the A. D. White Professor-at-Large program for the opportunity to come to Cornel1 and the members of the Baker Laboratory for their warm hospitality. We thank the Consiglio Nazionale delle Ricerche (CNR-Italy) for the award of postdoctoral fellowships for D.M.P. and the National Science Foundation for its support through Grant CHE-8912070. A knowledgeable reviewer and D. G.Truhlar made some very useful comments on our work. (77) Persky, A.; Broida, M. J. Chem. Phys. 1984, 81, 4352.

(78) Gordon, M. S.; Baldridge, K. K.; Bernhold, D. E.; Bartlett, R. J. Chem. Phys. Lett. 1989, 158, 189. (79) Hoffmann, R. Science 1981. 211, 995 and reference therein. (80) Elkind, J. L.; Armentrout, P. B. J. Phys. Chem. 1987, 91, 2037.